Antitumor Anthraquinones from an Easter Island Sea Anemone: Animal or Bacterial Origin?
2. Results and Discussion 1. Sea Anemone Dereplication
Samples of the sea anemoneGyractis sesere[26], also known asActiniogeton rapanuiensis[27], were collected in the intertidal zone of Easter Island, South Pacific Ocean, and extracted with chloroform to give 8 mg of a yellowish residue. The HRLCMS analysis of this crude extract showed the presence of two previously identified antitumor anthraquinones (Figure1) [24,25], Lupinacidin A (1) ([M + H]+m/z 341.1378) and Galvaquinone B (2) ([M + H]+m/z 369.3510), which were confirmed by complete NMR-spectroscopic characterization.
Figure 1.Identified molecules from the Easter Island sea anemoneGyractis sesere.
166
Mar. Drugs2019,17, 154
Interestingly, the crude extract of the sea anemone (Figure2) shows five resonances above 10 ppm;
a region which is characteristic for hydroxyl protons. The resonances atδ14.18 and 12.96 ppm are assigned to the groups at C-1 and C-6 of Lupinacidin A (1), as their vicinity and consequential hydrogen bonding to the ketogroups slows down their exchange. Similarly, the resonances atδ13.49, 12.50, and 12.14 ppm originate from the three hydroxylprotons in Galvaquinone B (2).
Figure 2.Chemical analysis of the crude extract of the sea anemoneGyractis sesere. (A)1H NMR spectra of the crude extract of marine anemoneGyractis sesereacquired in CDCl3, 600 MHz. Highlighted in the zoomed area are the frequencies of characteristic resonances originating from hydroxyl exchangeable protons in vicinity to ketogroups. (B) UV chromatogram (254 nm) of the crude extract of the sea anemoneGyractis seserehighlighting the specific peaks forRT: 24.3 min, and: RT: 25.2 min. (C) High resolution mass for(m/z [M + H]+341.1378) and (D) high resolution mass for(2) (m/z [M + H]+369.3510). *RT: Retention Time.
Other main peaks found in the sea anemone crude extract were peaks at RT 14.2 min with a HRMS [M + H]+m/z 295.19009 and at RT 23 min with a HRMS [M + H]+m/z 256.26312. Both exact masses were evaluated using the MarinLit database, however their HRMS did not match any known compound to-date.
167
Mar. Drugs2019,17, 154
2.2. Bacterial Metabolites and Harbored Bacteria
Lupinacidin A (1) and Galvaquinone B (2) have so far only been characterized in actinobacterial representatives, specifically from the generaStreptomyces[24,28] andMicromonospora[25], raising the question of the origin of these compounds in the sea anemone extract. Thus, we cultivated the Actinobacteria harbored by this sea anemone to determine if the anthraquinone producer was a bacterium or the sea anemone. Isolation media and the respective obtained strains are specified in Supplementary Table S1. Ten strains were identified through analysis of the 16S rRNA gene sequences as members of the generaMicromonospora,Streptomyces,Verrucosispora,Dietzia,Arthrobacter,Rhodococcus, andCellulosimicrobium(Figure3). Remarkably,Gyractis sesereharbors a high number of Actinobacteria genera, in total seven; the most abundant genus beingMicromonosporawith three different species, followed byArthrobacterwith two different species. Other actinobacterial genera were present with only one species each. Outstandingly, the only isolate belonging to theVerrucosisporagenus was the most abundant single Actinobacterium in the sea anemone.
Figure 3.Genera and number of Actinobacteria species strains isolated from the sea anemoneGyractis sesere.
2.3. Bacterial Growth
To evaluate the production of the anthraquinones by the isolated bacteria, we selected the actinobacterial representatives for growth experiments that were most closely related to known producers of the generaStreptomyces,MicromonosporaandVerrucosispora. In addition, we also grew the Dietziarepresentative due to the lack of comprehensive information about its secondary metabolite production. On the other hand,Rhodococcus,Arthrobacter, andCellulosimicrobiumwere omitted here because of their known poor production of secondary metabolites.
The growth yield of the selected Actinobacteria was in the range of 20 to 100 mg crude extract.
The chromatograms of HPLC analyses of the crude extracts were compared in order to facilitate the metabolic comparison between grown bacteria, the sea anemone and the pure substances (Figure4).
168
Mar. Drugs2019,17, 154
Figure 4.HPLC chromatograms of the crude extracts of the sea anemoneGyractis sesere, its respective actinobacterial isolates, and the purified anthraquinones, Lupinacidin A (1) and Galvaquinone B (2).
Approximate retention times of Lupinacidin A (1) and Galvaquinone B (2) are highlighted by boxes.
By comparison of chromatograms and retention times, we observed that Lupinacidin A (1) and Galvaquinone B (2) were only present in the crude extract of the sea anemoneGyractis sesereand Verrucosispora sp. SN26_14.1, and not in the other actinobacterial representatives. Clearly,Verrucosispora must be the producer of the anthraquinones. Further, it is obvious that the metabolites ofVerrucosispora sp.strain SN26_14.1 are dominant in the marine invertebrate. The chromatograms ofVerrucosispora and sea anemone extracts are nearly identical and differ only slightly in the region of retention time 20–23 min. Notably the chromatogram of the sea anemone extract also does not show any peaks that suggest the presence of metabolites of any other of the cultivated bacteria. Together, this strongly suggestsVerrucosisporaappeared to be the most abundant microbe in the sea anemone biomass during the collection.
The subtle difference in the metabolite profiles betweenVerrucosisporaand sea anemone extract in the retention time region 20–23 min appears to be to metabolites produced by the sea anemone itself.
Overall, the amount appears to be surprisingly small. This may however be caused by the isolation methodology (chloroform extraction), that prioritizes lipophilic substances and selects against the isolation of polar compounds such as peptides.
2.4. Actinobacterial Producer
To confirm and replicate the production of these metabolites, we undertook a scale up culture of Verrucosisporasp. SN26_14.1. Thus, 10 L of the Actinobacterium culture were grown, and extracted through the use of amberlite XAD-16 resin, yielding 1 g of crude extract with a brownish coloration.
This extract was subjected to stepwise flash chromatography using iso-octane and ethyl acetate gradients, which produced a total of ten fractions. The fractions were evaluated through HPLC to find the fractions containing Lupinacidin A (1) and Galvaquinone B (2). The chromatogram evaluation showed that only the orange colored fraction two, which was eluted with 90% iso-octane and 10%
ethyl acetate, contained 78 mg of metabolites enriched with Lupinacidin A (1) and Galvaquinone B (2). The purification of Lupinacidin A (1) and Galvaquinone B (2) was achieved through HPLC using normal and reverse phase chromatography.
169
Mar. Drugs2019,17, 154
Lupinacidin A (1) was isolated as a yellow powder, with a yield of 11 mg from a 10 L culture, which suggested to be an intermediate yield compared with Streptomyces and Micromonospora producers [24,25,28]. High resolution APCI-MS gave an [M + H]+adduct of m/z 341.1378, which results in a molecular formula of C20H20O5. The calculation of the degree of unsaturation indicated 11 degrees.1H NMR showed the characteristic exchangeable protons of (1) atδat 14.18 and 12.96 ppm, in addition to the three neighboring aromatic proton signalsδ7.26, 7.62, and 7.79 ppm that showed the expected coupling pattern for three neighboring aromatic protons in a para-ortho, ortho-meta relationship (two duplets, and one duplet of duplets). The13C NMR experiment showed 20 carbons of which two represented ketone signals (δ190.2 and 186.9 ppm), 12 aromatic carbons, and six aliphatic carbons (see Table1). Two-dimensional NMR experiments, Homonuclear COrrelated SpectroscopY (COSY), Heteronuclear Single Quantum Correlation (HSQC), and Heteronuclear Multiple Bond Correlation (HMBC), helped to confirm the identity of the molecules. These data were in agreement with the published information [24,25,28].
Table 1.Spectroscopic NMR data of Lupinacidin A (1) and Galvaquinone B (2).
Lupinacidin A
Galvaquinone B Position δC δHMult
(J in Hz) HMBC COSY δC δHMult
(J in Hz) HMBC COSY
1 162.5 157.5
2 117.5 137.1
3 159.6 141.1
4 130.4 153.8
4a 127.9 116.2
5 190.2 190.7
5a 117.1 116.2
6 162.6 162.7
7 124.3 7.26, d
(8.2) 5a, 6, 9 H-8 124.8 7.32, d
(8.5, 1.3) 5a, 6, 9 H-8
8 136 7.62, dd
(8.2, 7.5) 6, 9a H-7, H-9 137.1 7.72, dd
(8.5, 7.6) 6, 9, 9a H-7, H-9
9 118.3 7.79, d
(7.5) 5a, 7, 10 H-8 119.7 7.90, d
(7.6, 1.3) 5a, 7, 10, H-9
9a 133 133.4
10 186.9 186.5
10a 110.8 111.7
11 8.4 2.27, s 1, 2, 3 13.2 2.25, s 1, 2, 3
12 24.8 3.21, m
(6.6)
3, 4a, 13,
14 H-13 204.9
13 37.7 1.46, m
(6.6)
4a, 12, 14, 15
H-12,
H-14 42.4 2.85, m 12, 14, 15 H-14
14 28.4 1.80, m
(6.6)
12, 13, 15, 16
H-13, H-15, H-16
31.9 1.63, m 14, 15, 16,
17 H-15
15 22.5 1.04, d
(6.6) 13, 14, 16 H-14 27.6 1.63, m 14, 15, 16, 17
H-13, H-16, H-17
16 22.5 1.04, d
(6.6) 13, 14, 15 H-14 22.4 0.93, d,
(6.2) 14, 15, 17 H-15
17 22.4 0.93, d,
(6.2) 14, 15, 16 H-15
1-OH 14.18 1, 2, 10a, 13.49, s 1, 2, 10a
3-OH 5.62 2, 4a, 3
4-OH 12.50, s 3, 4, 10a
6-OH 12.96 5a, 6, 7 12.14, s 5a, 6, 8
**1H NMR (600 MHz) Solvent: CDCl3(δ1H, mult, J in Hz), ***13C NMR (125 MHz), Solvent: CDCl3.
Galvaquinone B (2) was isolated as a red powder, with a yield of 7 mg from a 10 L culture, which is an intermediate yield compared withStreptomycesandMicromonosporaproducers [24,28]. High
170
Mar. Drugs2019,17, 154
resolution APCI-MS gave an [M + H]+adduct of m/z 369.3510 and a molecular formula of C21H20O6. The calculation of the degree of unsaturation indicated 12 degrees. Galvaquinone B (2) showed characteristic exchangeable proton signals atδ13.49, 12.50, and 12.14 ppm, respectively. Aromatic signals were similar to those found in compound (1), showing three neighboring aromatic protons in a para-ortho, ortho-meta relationship (two duplets, and one duplet of duplets), but with a higher frequency (see Table1). The13C NMR experiment showed 21 carbons of which three represented ketone signals (δ205, 190.2 and 186.9 ppm), 12 aromatic carbons, and six aliphatic carbons (see Table1).
Two dimensional experiments (COSY, HSQC, HMBC) confirmed the identity of the molecules and were in agreement with the published data [24,28].
2.5. Phylogeny of the Producer
To determine whether the present isolate was a new species and to evaluate its evolutionary relationship, we performed a phylogenetic evaluation of the strain based of the 16S rRNA gene sequence (Figure5). The evaluation of the 16S gene showed a high similarity (99%) to the next related type strain,Verrucosispora marisDSM 45365T. However, analysis of the gyrase subunit B taxonomic marker (gyrB) showed 94.4% similarity toVerrucosispora marisDSM 45365T. The construction of the phylogenetic tree with the closest relatives in terms of the 16S rRNA gene sequence as well as known producers of (1) and (2) confirmed that the producer strain SN26_14.1 belongs to the genus Verrucosispora, and that quite likely it represents a new species within the genus. This result represents the first report of anthraquinone production for theVerrucosisporagenus. Interestingly, it appears that anthraquinones are more widespread metabolites in theActinobacteriaphylum, since compounds (1) and (2) have now been found in three actinobacterial genera,Micromonospora,Streptomyces, and Verrucosisporaproducers [24,25,28].
Figure 5.Phylogenetic tree based on 16S rRNA gene sequence ofVerrucosispora sp. SN26_14.1. The tree was calculated using a neighbor-joining statistical method and Jukes–Cantor model.•Red dots highlight Lupinacidin A (1) and Galvaquinone B producers (2).
171
Mar. Drugs2019,17, 154
2.6. Biosynthesis
Recently, the biosynthetic machinery for the production of compound (1) and (2) was described as a type II polyketide synthase (PKS) that features a special Baeyer−Villiger type rearrangement, and was allocated to an Rsd gene cluster inStreptomyces olivaceusSCSIO T05 [28]. The Rsd biosynthetic gene cluster (BGC) showed great similarity to the Rsl BGC reported for the production of rishirilide A and B inStreptomyces bottropensis(also known asStreptomyces. sp. Gç C4/4) [29]. The BGC Rsd is responsible for the production of six molecules (rishirilide B, rishirilide C, Lupinacidin A (1), Lupinacidin D, Galvaquinone A and Galvaquinone B (2)), and among them compound (1) and (2) [28]. This raised the question of whether inVerrucosispora,compounds (1) and (2) follow the same biochemical assembly line as described forStreptomyces. Thus, the genome ofVerrucosispora sp.SN26_14.1 was sequenced using Illumina MiSeq. Although the obtained short reads were not complemented with a long read sequencing technology as PacBio, we were still able to obtain a 6.9 Mb draft genome (NCBI Bioproject Access # PRJNA522941). This data was annotated with Prokka and analyzed with the Antismash online platform [30] to identify the secondary metabolite biosynthesis gene clusters. As shown in Figure6, the draft genome ofVerrucosisporasp. SN26_14.1 shared 60% of the genes of the Rsd gene cluster, as well as to an important percentage of the genes of the Rsl BGC. The similarity of the found genes ranged from 49% to 81%. The genetic architecture found in Vex BGC was quite similar to that of the Rsd and Rsl BGC. Remarkably, we could not detect any cyclase/aromatase and amidohydrolase sequences in our draft genome. Likely, this relates to the incompleteness of our sequence. Finally, it appears reasonable thatVerrucosisporasp. SN26_14.1 follows the same biosynthetic machinery for the production of Lupinacidin A (1) and Galvaquinone B (2) as found forStreptomycesspecies (Figure6).
Figure 6.Biosynthetic gene cluster of anthraquinones producers. Rsd:Streptomyces olivaceusSCSIO T05 gene cluster [28], Rsl: Streptomyces bottropensis(Streptomyces. sp. Gc C4/4) gene cluster [29], Vex:Verrucosisporasp. SN26_14.1.C1: aromatase,K1: acyl carrier protein,K2: ketosynthase (beta), K3: ketosynthase (alpha),A: acyl transferase,K4: 3-oxoacyl-ACP synthase III,T1: ABC-transporter (substrate binding), T2: ABC-transporter (ATP binding),T3: ABC-transporter trans-membrane, O1: luciferase-like monooxygenase,O2: flavin reductase,P: phosphotransferase,R1: SARP family regulator,C2: second ring cyclase,O3: 3-oxoacyl-ACP reductase,O4: anthrone monooxygenase,O5: NADH: flavin oxidoreductase,C3: cyclase,R2: SARP regulatory protein,R3: LAL-family regulator, O6: luciferase-like monooxygenase,R4: MarR family transcriptional regulator,T4: drug resistance transporter,O7: putative NADPH quinone reductase,O8: putative NADPH: quinone oxidoreductase, O9: FAD-dependent oxidoreductase,O10: C9-keto reductase,H: amidohydrolase,-3: unknown function, -2: major facilitator superfamily protein,-1: Transcriptional regulatory protein,1: cupin,2:
citrate/H+symporter,3: transcriptional regulator.
2.7. Antibiotic Activity Test
We performed a disc diffusion antibiotic test as a preliminary evaluation to determine if Lupinacidin A (1) and Galvaquinone B (2) have an inhibition effect on bacteria. As a positive control,
172
Mar. Drugs2019,17, 154
we used streptomycin at a concentration of 25μg/disc. The results showed that Lupinacidin A (1) and Galvaquinone B (2) did not produce any growth inhibition against the Gram-positive bacterium Staphylococcus lentusDSM 20352T, and neither against the Gram-negative bacteriumEscherichia coli DSM 498T. In contrast, the positive control, streptomycin produced an inhibition halo of 22 mm for Gram-negative and 18 mm for Gram-positive bacteria.
3. Materials and Methods